Self-awareness of biases in time perception☆
Introduction
How did it get so late so soon? – Dr. Seuss
The literature in psychophysics documents how we make time evaluations and demonstrates that our subjective assessments of the time that passes may be inaccurate. When individuals are explicitly given time targets, discrepancies between objective time and subjective time are observed as a function of the time duration range – below or above 1 s –, but also the specific task design (Grondin, 2010, Grondin, 2014, Wearden, Lejeune, 2008). Furthermore, emotional states (Droit-Volet, Meck, 2007, Droit-Volet, Wearden, 2002, Fayolle, Gil, Droit-Volet, 2015), attentional constraints (Buhusi and Meck, 2006), drug usage (Wittmann et al., 2007) and incentives to be accurate (Akdogan, Balci, 2016, Çoşkun, Sayali, Gürbüz, Balci, 2015) also shape our perception of time and perception accuracy. Even though the mechanisms of time-keeping are reasonably well understood, little is known about how aware people are of their potential biases and whether this knowledge affects their behavior when facing time-related incentives.
This question is related to our ability of being aware that we know something, a concept referred to as metacognition (Koriat, 2007, Nelson, 1996). Metacognition has been applied to subjective time only recently (Akdogan, Balci, 2017, Lamotte, Izaute, Droit-Volet, 2012, Sackett, Meyvis, Nelson, Converse, Sackett, 2010) to show that individuals hold beliefs about distortions in their experience of time that affect their time-related judgments. However, it is unclear how such beliefs impact decision making. For instance, if a person is typically late, is she bound to fail to meet important deadlines? Or is she conscious enough of her tendency to be able to correct it when the incentives are sufficiently high?
The objective of this study is to embed a classical time production paradigm into a decision-making paradigm, where participants are not given explicit time targets but, instead, choose their time targets as a response to varying incentives. With this approach, we can investigate the ability of individuals to internalize their perceived biases in their decision of when to act. This study complements the growing literature that links timing to decision-making. However, instead of investigating timing as a decision-making process itself as in traditional accumulator timing models (Gibbon, 1977, Treisman, 1963) or in studies looking into the processing dynamics underlying temporal decisions (Balci and Simen, 2014), we investigate the decisions that result from timing. Said differently, we model timing properties as an input for decision-making rather than an output of decision-making.1
Our study faced four main challenges. First, to isolate biases in time-keeping behavior, it was crucial to choose a task that relied as little as possible on orthogonal concerns. To prevent interference with memory processes, we avoided retrospective paradigms (Block, Zakay, 1997, Fraisse, 1984) and we opted for a prospective paradigm in which participants were informed beforehand that they had to make a time related judgment. We also favored time production (the length of the interval to produce is disclosed) to time reproduction (participants have to reproduce an interval of time of unknown length). Even though time production tasks require some form of memory regarding the length of the intervals involved in the experiment, they do not tax working memory to process and store information on the time interval to reproduce as time reproduction tasks do (Baudouin et al., 2006). We finally designed a novel task that discouraged chronometric counting (Grondin, Killeen, 2009, Hinton, Rao, 2004, Wearden, 2003) to better capture intrinsic timing properties.
Second, to be able to focus on how participants adjust their time targets to respond to incentives, it was critical to design a task in which we varied the structure of the rewards and paid participants as a function of their performance. This design departs from traditional experiments on subjective time. Indeed, with a few exceptions (Akdogan, Balci, 2016, Wearden, Grindrod, 2003) participants are usually asked to produce accurate reports and are compensated with a fixed payment for their effort. Incentivized payments are, however, crucial in decision-making studies because the objective is to trigger cognitive processes that govern real-life choices. Concretely, we included three tasks, each featuring specific time-related incentives. In our baseline task, participants completed a time production task in the range of 31–41 s, and they were rewarded for accuracy. Formally, they earned maximal payoff ($20) when they correctly estimated the announced time and their payoffs was reduced symmetrically as their estimates were farther away from the announced time in either direction. We then asked participants to complete two time production tasks in which the reward structure was altered. In one of them, the announced time was a hard deadline: earnings increased as the estimate came close to the announced time and vanished afterwards. This reward scheme incentivized participants to avoid exceeding the deadline and to report lower estimates compared to the baseline task. In the other one, the announced time was a release time: earnings were nil for estimates below the release time, maximal at the release time and decreased afterwards. This scheme incentivized participants to wait past the announced time and to report higher estimates compared to the baseline.
Third, to assess whether participants were aware of their intrinsic timing attitudes, we needed to elicit their beliefs. There is a variety of belief elicitation methods. Recent studies show that eliciting beliefs with properly incentivized methods are often meaningful and consistent with observed behavior in the laboratory (Palfrey, Wang, 2009, Schotter, Trevino, 2014). Furthermore, the experimenter should minimize the chances that the procedure itself affected the behavior under study. In particular, we were not interested in feedback effects, which are known to affect timing behavior (Brown, Newcomb, Kahrl, 1995, Droit-Volet, Izaute, 2005, Franssen, Vandierendonck, 2002). For these reasons, we opted for an incentivized method and a procedure that did not interfere with the main tasks. At the end of the experiment, we asked our participants to make a self-assessment regarding their performance in the baseline task and we rewarded them for accuracy.
Fourth, to be able to assess biases in time perception and to understand their effects on decision-making, we had to develop a normative theoretical framework that generated predictions which could then be compared to the empirical findings. In the tradition of neoclassical economics, we opted for a rational framework in which decision-makers maximize their expected payoffs but are subject to noisy time perception. More precisely, we modeled individuals as in traditional theories of subjective time (Gibbon et al., 1984) and we assumed that, when they target a given time interval, they produce time intervals on average equal to the target (“mean accuracy”). We derived the optimal behavior of such decision-maker in the context of our time-related incentives. We contrasted the predictions of the model with the actual behavior of participants and we determined whether deviations from the target intervals were due to biases in time perception or biases in decision-making.
Given the growing interest in the role of emotions on time perception (Droit-Volet, Meck, 2007, Droit-Volet, Wearden, 2002, Fayolle, Gil, Droit-Volet, 2015) and decision-making (Damasio, 1994, Phelps, Lempert, Sokol-Hessner, 2014, Schwartz, 2000), we were also interested in the effect of emotions on the interaction between the two. Emotion, however, is a complex and rich concept and we opted for an exploration of the effect of physiological stress on time-related incentives. To this purpose, half of our population completed the Cold Pressor Task (CPT), a method that increases the participants’ cortisol levels and has been previously shown to affect decision-making (Lighthall, Sakaki, Vasunilashorn, Nga, Somayajula, Chen, Samii, Mather, 2012, Porcelli, Delgado, 2009). In particular, in the context of psychological stress, increases in cortisol levels correlate with increases in dopamine release (Pruessner et al., 2004), which are known to make time perceived as passing more slowly (Meck, 1996). This evidence suggests a possible mechanism linking stress to changes in time perception (through variations in dopamine concentration).2 We therefore hypothesized that participants who completed the Cold Pressor Task would act as if time passed more slowly.
To answer our research questions, we followed a simple research strategy. We derived hypotheses for a rational decision-maker who maximizes rewards, and fitted predictions to the data. We used time reports in the baseline task to assess the presence of a time perception bias, for each individual and in the overall population. By comparing the responses between the baseline and each of the two other tasks, we assessed whether participants responded to incentives conditional on their biases. That is, we determined whether biases in behavior in those two tasks were due to biases in time perception or biases in decision-making. Moreover, by comparing behavior between participants who did and did not complete the CPT, we studied the effect of physiological stress on time perception and decision-making. Finally, we used elicited beliefs to evaluate self-awareness and to further ascertain if behavior was contingent on beliefs.
Section snippets
Experimental design and procedures
We conducted two treatments of a single experiment in the Los Angeles Behavioral Economics Laboratory (LABEL) at the University of Southern California, with a large number of subjects (170 individuals, 79 Male and 91 Female, in 21 sessions of 6–10 participants each). Among those, 83 individuals (36 Male and 47 Female) participated in the “Stress” treatment and underwent the Socially Evaluated Cold Pressure Task (CPT). Subjects completed a survey before entering the laboratory to ensure all
Theory and predictions
We start with a theoretical exercise to assess how participants who believe that their subjective perception may be noisy and biased should behave. This exercise allows us to make testable predictions and to identify in our data whether behavior is driven by the existence of a bias that is not corrected, by mistakes in decision-making or by both.
Results
The objective of the empirical analysis was to test Predictions 1–4. We do not report the analysis of each time interval separately in the main text as it is not central to our study. However, the analysis of each interval is briefly addressed in Appendix A.5. We also use a p-value of 0.05 as the benchmark threshold for statistical significance.
Behavioral adaptation model
If participants were at least partially aware of their biases, why were they not correcting better for them in tasks U and O? A possible explanation is that, even though they knew their tendency to misrepresent time in a given direction, they were unsure of the magnitude of their bias. Recall that an unbiased participant i should target t in task B, in task U and in task O. Recall also that the optimal shading should be symmetric around t: . Suppose that
Discussion
It has been proposed that perceived time synchronizes with the ticking of an internal clock (Treisman, 1963), an idea mathematically represented by the scalar timing model (Gibbon, 1977). This internal clock consists of a pacemaker that emits pulses (ticking) stored in an accumulator. Accuracy is obtained when the pacemaker is perfectly synchronized with objective time, and distortions occur when it speeds up or slows down. In our framework, when a time interval is announced, the participant
References (53)
- et al.
The effects of payoff manipulations on temporal bisection performance
Acta Psychol.
(2016) - et al.
Decision processes in temporal discrimination
Acta Psychol.
(2014) - et al.
Age-related changes in duration reproduction: involvement of working memory processe
Brain Cogn.
(2006) - et al.
Time estimation: does the reference memory mediate the effect of knowledge of results?
Acta Psychol.
(2002) - et al.
Scalar timing in memory
- et al.
Psychosocial versus physiological stress - meta-analyses on deactivations and activations of the neural correlates of stress reactions
Neuroimage
(2015) - et al.
Awareness of time distortions and its relation with time judgment: a metacognitive approach
Conscious. Cogn.
(2012) - et al.
Neuropsychological mechanisms of interval timing behavior
Bioessays
(2000) Time perception measurements in neuropsychology
- et al.
Emotion and decision making: multiple modulatory neural circuits
Annu. Rev. Neurosci.
(2014)
Acute stress modulates risk taking in financial decision making
Psychol. Sci.
Temporal discrimination and the indifference interval: implications for a model of the internal clock
Psychol. Memographs
Impaired time perception and motor timing in stimulant-dependent subjects
Drug Alcohol Depend.
Are you early or late? Temporal error monitoring
Journal of Experimental Psychology: General
Prospective and retrospective duration judgments: a meta-analytic review
Psychonomic Bull. Rev.
Stress and decision-making in humans: performance is related to cortisol reactivity, albeit differently in men and women
Psychoneuroendocrinology
The brain as a hierarchical organization
Am. Econ. Rev.
Stress induces contextual blindness in lotteries and coordination games
Front. Behav. Neurosci.
How Long is a Minute?
Temporal-signal detection and individual differences in timing
Perception
Interval timing with gaps and distracters: evaluation of the ambiguity, switch, and time-sharing hypotheses
J. Exp. Psychol.
Anomalous Time Preferences? The Role of Subjective Time Perception
Optimal time discrimination
Q. J. Exp. Psychol.
Descartes Error: Emotion, Reason, and the Human Brain
Acute stressors and cortisol responses: a theoretical integration and synthesis of laboratory research
Psychol. Bull.
The effect of feedback on timing in children and adults: the temporal generalization task
Q. J. Exp. Psychol. A
Cited by (14)
How time flies: Time perception and intertemporal choice
2024, Journal of Behavioral and Experimental EconomicsThe effect of feedback on temporal error monitoring and timing behavior
2019, Behavioural Brain ResearchCitation Excerpt :In a similar vein, for the signed feedback but not the absolute feedback group, global metacognitive judgments were correlated with the mean direction of actually committed errors, demonstrating a relatively larger awareness of these errors. These results contrast with those reported by Brocas et al. [7] who found that their participants were fairly good in judging whether they under- or overproduced temporal intervals. However, several differences in the method and the range of tested intervals make it difficult to compare these two studies.
Metric error monitoring in the numerical estimates
2019, Consciousness and CognitionCitation Excerpt :In the timing domain, Akdoğan and Balcı (2017) showed that in a temporal reproduction task using durations ranging between 1.5 and 6 s, participant's confidence ratings and error-directionality judgments reflected the amount and the direction (over- or under-reproduction) of errors on a trial-to-trial basis. Moreover, for longer durations, people are also aware of their general biases in time perception (Brocas, Carrillo, & Tarraso, 2018). For number representations, a few studies hinted at a possible numerical error monitoring mechanism (e.g., Gelman & Gallistel, 1978; Kibbe & Feigenson, 2015).
How long is a minute?
2018, Games and Economic BehaviorCitation Excerpt :Even though we cannot address this phenomenon with the data collected, we believe that a better understanding of the differences between corrected and uncorrected time perception is directly relevant to decision-making. In a follow-up research (Brocas et al., 2018) we investigate the issue of awareness of time perception biases. We show in an incentivized experiment that subjects are aware of their biases, but only partially, and they do not fully correct for them when producing time.
Metric error monitoring as a component of metacognitive processing
2024, European Journal of NeuroscienceEmersiology in Sport Science: The Unconscious Living Body in the Case of Corporeal Non-Property
2024, Sport, Ethics and Philosophy
- ☆
We are grateful to Andreas Aristidou, Calvin Leather and the members of the Los Angeles Behavioral Economics Laboratory (LABEL) for their insights and comments in the various phases of the project. We also acknowledge the financial support of the National Science Foundation grant SES-1425062.